shock detection changes seem to make hybrid Riemann worse
zingale opened this issue · comments
We are getting some artifacts now with the odd-even problem:
http://groot.astro.sunysb.edu/Castro/test-suite/gfortran/2024-01-21/oddeven-2d-hybrid.html
the issue seems to be that our "shock" tag is now 1 zone narrower (3 zones wide instead of 4):
(top is new, bottom is old)
This is something that can be adjusted by setting the threshold parameter
indeed, changing the threshold from the default 2/3 down to 1/2 seems to make the problem look good again:
(left is castro.shock_detection_threshold=0.66666, right is castro.shock_detection_threshold=0.5)
closing this, as it just seems to be a threshold issue